On 1/30/19 8:27 AM, Greg KH wrote: > On Fri, Jan 25, 2019 at 02:13:16PM +0000, Vladimir Murzin wrote: >> Some designs, like MPS3, expose number of virtual serial ports which >> already close or exceeds MPS2_MAX_PORTS. Increasing MPS2_MAX_PORTS >> would have negative impact (in terms of memory consumption) on tiny >> MPS2 platform which, in fact, has only one physically populated UART. >> >> Start with converting existent static port array to idr. As a bonus it >> make driver not to fail in case when no alias was specified in device >> tree. >> >> Signed-off-by: Vladimir Murzin <vladimir.murzin@xxxxxxx> >> --- >> drivers/tty/serial/mps2-uart.c | 47 ++++++++++++++++++++++++++++-------------- >> 1 file changed, 31 insertions(+), 16 deletions(-) >> >> diff --git a/drivers/tty/serial/mps2-uart.c b/drivers/tty/serial/mps2-uart.c >> index 9f8f637..6da0633 100644 >> --- a/drivers/tty/serial/mps2-uart.c >> +++ b/drivers/tty/serial/mps2-uart.c >> @@ -22,6 +22,7 @@ >> #include <linux/serial_core.h> >> #include <linux/tty_flip.h> >> #include <linux/types.h> >> +#include <linux/idr.h> >> >> #define SERIAL_NAME "ttyMPS" >> #define DRIVER_NAME "mps2-uart" >> @@ -397,7 +398,7 @@ static const struct uart_ops mps2_uart_pops = { >> .verify_port = mps2_uart_verify_port, >> }; >> >> -static struct mps2_uart_port mps2_uart_ports[MPS2_MAX_PORTS]; >> +static DEFINE_IDR(ports_idr); > > You forgot to call idr_destroy() when your code unloads :( Hmm, but code doesn't unload since ce87122911f8 ("serial: mps2-uart: make driver explicitly non-modular") or I'm missing something? Cheers Vladimir > > Yeah, it's not an obvious design pattern, I think someone will fix it up > someday... > > thanks, > > greg k-h >